@JoshuaMoelans, can you give this a first once over? This is not fully completed according to the scope I had in mind. Still, the PR reflects the most significant portion requiring restructuring, and I would appreciate some SDK internals WTF count from your POV.

I don't know what is going on with the cursor bot today, but the "bugs" it raises are pure gold:

#1413 (comment)
Why is the comment misleading? It specifies what the function assumes, and the "bug" description verifies (the client includes the null-termination in the length argument). And regarding the more critical (after the seemingly less critical aspect): let call-sites allocate means that this is a static helper, and the translation-unit call-sites are responsible for maintaining path_w. The call site and the static helper are written and used by the same role.

#1413 (comment)
What is confusing is the cursor bot text. It has a very long-winded way of saying in the "bug" description that the code does what it's supposed to. And then raises a concern with its taste buds.

#1413 (comment)
This is very intentional because assignments often trigger static analysis warnings when used in a condition.

And so densely written, it is a pleasure to decipher. This is clearly the future of software development 🤗

Bug: Crash Reporter: Redundant Conditional Compilation

The Windows version of the crash reporter uses wmain but the non-Windows version uses main. However, both versions end with #ifdef _WIN32 blocks that call _wremove(path) on Windows and remove(path) on non-Windows. The problem is that in the Windows wmain version, path is a wchar_t *, but in the non-Windows main version, path is a char *. This means the #ifdef _WIN32 block at line 100-104 is checking the wrong condition - it's inside a function that's only compiled on Windows (wmain), so the #ifdef _WIN32 is redundant and the #else branch will never execute. This appears to be a logic error in the conditional compilation structure.